With the advent of the Bureau of Justice Assistance’s Smart Policing Initiative, the increased use of the SARA model of problem-oriented policing, and the need to do more with less, agencies are more and more evaluating their own crime prevention initiatives. Many of these efforts have involved using experimental and quasi-experimental designs, when possible, to create more certainty about the believability of these evaluations.

But how do agencies conduct their own experiments, and what are some challenges in doing so? Building on the experience of the Sacramento Police Department and Sgt. Renee Mitchell’s efforts, this demonstration provides a step by step guide on how agencies can conduct their own experimental evaluations. The guide will not only include information on the science of experiments (e.g. how to design an experiment, the statistical benefits of experiments, how and why randomly allocating units is useful), but also a discussion of the prospects and pitfalls for conducting experiments within police agencies. In particular, the guide will focus on addressing potential challenges to agency-led randomized trials. These include using training as a means to teach officers about the value of experiments, identifying change agents in the department and getting them on board to increase officer buy-in, and working with department management to ensure that top leaders are fully committed to the study and are ready to make use of the results.

This guide will be an important addition to existing publications and tools on conducting evaluations more generally, as it specifically focuses on experimental evaluations.